Famine amid plenty

Why can people starve in a year when their region harvested enough food to feed everyone?

During the Ethiopian famine of 1973, food moved out of Wollo, the province where people were dying, and into provinces where they were not. No blockade forced it. No one was hoarding it out of malice. Lorries carried grain away from starving people, on ordinary commercial terms, because that is where the grain was worth more.

Once you have sat with that sentence for a moment, the usual framing of famine starts to look wrong. We ask "was there enough food?" and treat the answer as decisive. But Wollo suggests the question does not settle anything. What question would?

Try the obvious defence first: perhaps the food left because there was more elsewhere and prices there were higher. Not so. Grain moves toward the higher price, and a price is not a measure of need — it is a measure of what somebody can and will pay. A person with no income, however hungry, bids nothing. Their hunger is not a weak signal in the market. It is not a signal at all.

So food does not flow down a gradient of hunger. It flows down a gradient of purchasing power, and in a famine those two gradients point in opposite directions — because the same event that empties a person's stomach usually empties their wallet first.

That reframing is Amartya Sen's, and it is worth stating in his terms. What matters for a given person is not the regional total but what they can lawfully obtain. Sen calls the ingredients an endowment — what you actually hold: your labour power, your plot, your animals, your savings — and an exchange mapping — the terms on which the world will convert those into food, meaning wages, employment, and the price of grain against whatever you sell. Run one through the other and you get an entitlement set: the bundles genuinely available to you. Starvation is that set failing to contain any bundle with enough food in it.

Now the puzzle dissolves into two clean routes, neither of which needs the harvest to fail.

The endowment can collapse. In Bangladesh in 1974, floods destroyed the transplanting and weeding work months before they threatened any crop. Landless labourers lost wages immediately; the harvest they were worrying about was still in the future, and food availability per head that year was in fact near a local high. The famine arrived ahead of the shortage.

Or the exchange terms can turn. In Wollo, pastoralists watched their animals lose value against grain far faster than grain rose — a herd that fed a family last year bought a fraction of the food this year. Nothing about the herd changed. The rate did.

Bengal in 1943 is the famous case, and I should be careful with it. Sen argued that aggregate food availability was not exceptionally low, and that the killing mechanism was wartime inflation and a boom in Calcutta bidding rice away from rural labourers whose wages did not move. Something on the order of two to three million people died, though the estimates are genuinely disputed. So is Sen's premise: later historians have argued the 1942 cyclone and a fungal disease of the aman crop cut the harvest more than he allowed, so a real supply shortfall was in play alongside the entitlement collapse. The entitlement framework does not need Bengal to have had normal supply; it only needs distribution to be doing work that the totals cannot see.

Two refinements, and one honest limit.

The first is that entitlement collapse is self-reinforcing. Once traders expect scarcity they withhold stock, which raises the price, which shrinks entitlements further, which converts a modest shortfall into a rout. This is why famines can outrun their causes, and why the classic policy response is not to ship food but to hand out cash or wages: restore purchasing power in the famine district and the same lorries turn around, because now the gradient points inward.

The second is that "enough food existed" is not an accusation of cruelty. Nobody in the Wollo lorries decided anything. The mechanism is entirely impersonal, which is what makes it hard to see and hard to blame.

The limit: Sen's framework explains who starves and why, but it does not follow that supply never matters — a genuine collapse in output will cause famine even under perfectly equal distribution, and critics reasonably argue that some famines Sen read as distributional had a larger supply component than he allowed. His further claim, that famine does not occur in a functioning democracy with a free press, is likewise contested, and rests on a demanding reading of "functioning".

• Why cash relief often pulls food into a famine district faster than shipping grain does.
• How speculative withholding turns a small expected shortfall into a large realised one.

Every term the collection defines is gathered in the glossary.